Uncertainty in modeling fitting and simulations of a continuously stirred tank reactor (CSTR)

Abstract

Continuously stirred tank reactors (CSTRs) play a big role in chemical engineering and in industries nowadays. Therefore, to produce quality products, it is important and necessary to make sure CSTRs work successfully by avoiding disturbance from external factors. In this work, we look at the impact of uncertainties (errors) to the model fitting and the parameter estimation of the CSTR, for both cases of exothermic and endothermic reactors. We introduced different types of noise (random noise, quasi random noise and systematic random noise) with different levels, low and high variance in the numerical solutions to generate the noisy data. Two parameters, activation energy E and reaction rate k have been estimated and investigated, due to their sensitivities to the system facilitated by the guessed noisy data and by using least square method. The results showed that, the increase of noise in measurements affects activation energy E much more than the reaction rate k for all types of noise as k converges to its true values. Also, from the results obtained, the endothermic CSTR is more sensitive than the exothermic CSTR for any kind of noise according to much deviation observed in its estimated activation energy values. Furthermore, quasi random noise has provided closer values for parameter estimation compared to the other types of noise but with negative activation energy for both cases (exothermic and endothermic). Keywords: CSTR, Exothermic, Endothermic, Random noise, Systematic random noise and Quasi random noise.